International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169

Volume: 4 Issue: 12 01 04
_______________________________________________________________________________________________
Analysis of Load Balancing Technique using Clustering Approach by PSO
Algorithm

Reshami Pathekar Prof. A. B. Jirapure

M. Tech. Dept. of E&TC Dept. of E&TC
Priyadarshini College of Engineering Priyadarshini College of Engineering
Nagpur, India Nagpur, India
reshmi.pathekar01@gmail.com Ashish_ji2014@rediffmail.com
.

Abstract--Wireless sensor Networks made up of nodes, where each node is connected to one or more number of sensors. WSN consists of a
large number of low-cost, low-power, multifunctional sensor nodes. The life of sensors depends on their energy. When sensors in the network
communicate with each other their energy get lowered and after some time they may die and because of that the network fails. Therefore our aim
is to reduce the energy consumption and increase the sensor network lifetime. As the energy available in the sensor nodes used in WSN is
limited, therefore we are trying to propose the technique to maximize lifetime of WSN by PSO algorithm. In this paper the selection of cluster
head is done via PSO algorithm. Particle Swarm Optimization (PSO) is the technique used for placing the sink node. The sink node is at optimal
position. The cluster head is selected on the basis of energy of the clusters and connectivity among the clusters. The cluster head then
communicate with the Base Station (BS).
Index Terms: WSN, PSO, BS, AN.

__________________________________________________*****_________________________________________________

I. INTRODUCTION determines the lifetime if the sensors. The data collected

from wireless sensor networks is usually stored in the form
In WSN to detect the changes in the environment
of numerical data. Data is stored in a central base station.
such as temperature and the movement vibration etc. the
WSN has a wide range of applications which are useful to
coverage area is needed for wireless sensor networking. The
industry, science, transportation, civil infrastructure, and
wireless sensors are very significant due to the various
security. In WSN the nodes can communicate with each
resource constraints. The resource constraints such as the
other or with the cluster head. While communicating with
detection range of sensors, the number of sensors and the
each other the energy of the sensors get lowered. When the
environmental condition. The optical deployment of
energy of sensors get lowered, they may die and the network
wireless sensors and radars is done on the above constraints.
fails. So it is important to increase the lifetime of WSN.
Due to these constraints, it is very important to design an
energy-efficient scheme. An energy-efficient scheme that In this paper, we propose the method for maximize
consumes energy uniformly across the sensing field to the lifetime of WSN using clustering approach via PSO
increase the network lifetime. The agriculture, target algorithm. The cluster head is selected by the PSO
tracking, health care monitoring system etc. are the various algorithm. PSO algorithm is used to find out the optimal
application of WSN. In WSN the sensors position of the sink node. Then cluster head is selected by
usingthe sink node position.Clustering is useful to reduce
are of low costand also has a low power. When sensors
the collisions between the sensors. It is also useful in
loses their energy there is no guarantee of connectivity of
balancing load among the sensors. By selecting the cluster
network and the coverage in wireless sensor network. When
head via PSO algorithm, the network lifetime increases as
the battery of the sensor get reduced, its not easy to change
the communication is done between the cluster head and the
or to replace the battery of the sensor. In the critical
base station. Cluster head collects the data fromthesensors
applications such as forest, border surveillance applications,
within the cluster and then transfer it to the base station.
etc. its not easy to replace the batteries. One major problem
in a WSN is to produce low cost sensors and the size of the II. LITERATURE REVIEW
sensor nodes should be small. WSNs may be useful in
Particle swarm optimization (PSO) is an algorithm
various environments, such as remote and hostile regions on
depends upon swarm of particles. The Application Node
large scale. In WSN, the ad hoc communications are a key
(AN) position is assumed in traditional approach and uses
component. Energy and Power Consumption should be
PSO algorithm to find out the optimal position of the sink
minimized which is useful for sensing device. Sensor nodes
node. The positions of ANare not pre-determined in a real
should be energy efficient. Limited energy resource
1
IJRITCC | December 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication
Volume: 4 Issue: 12
_______________________________________________________________________________________________
time therefore this approach is ineffective. The positions of
the Application node are manually determined. PSO
algorithm used for optimal placement of base station. The
paper is proposed by B. Paul et al. [4]. It states that PSO is
used to reduce the power consumption and maximize the
network lifetime of two tired wireless sensor network. PSO
algorithm is used to find out optimal position of sink node.
A cluster formation scheme which is named as LEACH is
proposed by Heinzelmanet al. [5] this protocol gives the
smallest number of clusters this scheme assumes that the
energy consumption of cluster head should be uniform and it
does not guarantee the good cluster head distribution.
Younis and Fahmy proposed HEED protocol [6]. In this
scheme the cluster head selection is depend upon the
residual energy of the sensors and cost of the sensors. The
properly distributed cluster head are produced by this
scheme. It also produced the compact cluster. In lossy
wireless sensor network Gong et al. [7] proposed energy
efficient clustering scheme which is based on link quality.
ISSN: 2321-8169
01 04
one cluster head inside each cluster. If there is no neighbor
within the range of the sensor then the sensor itself is the
cluster head [1]. Otherwise the selection of cluster head
depends on the percentage of residual energy. The sensors
are sorted on the basis of residual energy and the
connectivity among the other neighbors. Then the cluster
head is selected. When the energy of cluster head get
reduced the iteration takes place and sensors updates their
local information.
Only one cluster head can be selected in each
cluster by using PSO algorithm. Cluster head is selected on
the basis of range, connectivity among the clusters and the
residual energy. Then the cluster head will communicate
with the other clusters within the range and then
communicate with the base station. When the energy of
Cluster head reduces, then again run the PSO algorithm and
find the cluster head.

III. PROPOSED WORK

In the Proposed approach application node
selection is made dynamic during the simulation of PSO
algorithm. Particle swarm optimization (PSO) is a
optimization technique which is population based. It is
inspired by social behavior of bird flocking or fish
schooling. These particles are moved around in the search-
space. The improved position guide the swarm movement. It
is known as the entire swarm's best known position. The
process is repeated. As compared to other methods PSO gets
better results. Sensor node clustering is the techniques that
can expand the lifespan of the whole network at the cluster
head through data aggregation. In this paper, the clustering
is done using particle swarm optimization (PSO) algorithm
for wireless sensor networks. This maximizes the lifetime of
the network. The application node position plays an Fig. 1 Sensor communicating and dropping their energy
important role in deciding the efficiency of Particle Swarm From fig.1 as the energy level decreases the color of nodes
Optimization. The PSO algorithm is used for optimized BS changes from Green to Yellow. Again when the energy
locations to reduce power consumption and maximize the level decreases the color of nodes changes from Yellow to
life time of WSN. After finding optimized location for sink Red. The Red color of nodes indicates that the node is
node, The Sink node transfer data to application node (AN) dead.Selection of the CH is done on the basis of
and AN to regular nodes. Regular nodes are attached to AN. connectivity among the clusters. Range of the CH is defined
Those nodes which have same values send response to the for the communication in the cluster.
application node via sink node. The initial energy of the
nodes are set randomly and the sink is assumed to have CH sends request to all the sensor nodes within the cluster
infinite energy. The transmission and reception of data by range.The other nodes send their current energy status.
the nodes is assumed to be Omni-directional. Higher residual energy nodes will be identified. The cluster
members communicate with the CH and send a data. The
Basically the selection of cluster head is the nodes which are not within the cluster head range can
essential operation in clustering. To increase the lifetime of communicate through the other nodes. CH will collect the
network the cluster head selection is on the basis of higher data from all the nodes within the cluster. CH will transmit
residual energy. For clustering the first priority is the the collected data to the base station. If the residual energy
percentage of residual energy of each sensor. There must be of the CH is low or the load on the CH increases, the new
2
IJRITCC | December 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication
Volume: 4 Issue: 12
_______________________________________________________________________________________________
CH is selected. The load is shifted to the new CH. The new
CH collects the information from the other nodes and send it
to the base station. The old CH will act as the general node.
IV. PSO ALGORITHM
Input the number of nodes.
Calculate the position of the nodes.
Calculate the probability of CH is to be selected.
Evaluate the fitness value for each node.
Calculate the best fitness value.
Calculate the particle velocity.
Find the range of the sensor node.
Run the PSO algorithm.
Update velocity of the particle.
Update the position of the particle.
LOAD BALANCING ALGORITHM
Initialise the global variables.
Initialise the nodes as Set now [$ns now]
Calculate the position of the nodesusing the rand
function.
Calculate the node 1 velocity
Calculate the node 2 velocity
Compare the velocity of both the nodes
If { node 1 = node 2 }
Then Load is shifted from node 1 to node 2
V. SOFTWARE BEING USED
Network Simulator (NS2) is used for the simulation
of proposed method.
VI. SIMULATION RESULTS
The comparative analysis of clustering approach
using PSO algorithm with load balancing and clustering
approach using PSO algorithm without load balancing
technique is given below.
Fig. 2 Residual Energy of clustering with load balancing
andclustering without load balancing in Joules
IJRITCC | December 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
ISSN: 2321-8169
01 04
The residual energy of the clustering with PSO and
load balancing technique and without load balancing is
calculated on the NS2 simulation. During the simulation in
NS2 every time we need to change the no. of nodes and
evaluate the respective residual energy level. Hence from
the figure it is observed that initially the residual energy
using the load balancing approach is 0.2J and then it is
decreases to 0.04J. In without load balancing approach
initially it is 0.2J and then it is nearly equal to zero. As we
can see from the graph the residual energy decreases
because as the no. of nodes increases, the communication
between the nodes require more energy. Residual energy in
load balancing is more than without load balancing.
Therefore we can say that load balancing technique is
efficient to maximize the life time of WSN.
During the simulation in NS2 every time we need
to change the no. of nodes and evaluate the respective
throughput. Hence from the figure it is observed that
initially the throughput using the load balancing approach is
30 bits/sec and then it is increases 120 bits/sec. In without
load balancing approach initially it is 28 bits/sec and then it
is 102 bits/sec
Fig. 3 Throughput of clustering with load balancing
andclustering without load balancing in Bits/Sec
Fig. 4 PDR of clustering with load balancing andclustering
without load balancing
3
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 4 Issue: 12 01 04
_______________________________________________________________________________________________
From the figure it is observed that initially PDR
using the load balancing approach is 110 and then it is
increases 450. In without load balancing approach initially it
is 100 and then it is 400.

VII. CONCLUSION
There are many ways to extend the life time of the
WSN application. One such method is particle swarm
optimization. In this paper, existing solutionsfor finding out
optimized location for sink node using PSO is further used
to find out the cluster head. The PSO algorithm is used to
achieve good scalability, to reduce the energy consumption
and increase network lifetime. It is useful to shorten data
transmission.
REFERENCES
[1] Miao Zhao, Yuanyuan Yang and Cong Wang, Mobile data
gathering with load balance clustering and dual data
uploading in wireless sensor networks, IEEE Trans. Mobile
Comput., vol. 14, no. 4, April 2015.
[2] R. Alageswaran, R. Usha, R. Gayathridevi and G. Kiruthika
Design and Implementation of Dynamic Sink Node
Placement Using Particle Swarm Optimization For Life Time
Maximization of WSN Applications, IEEE-Internationa
Conference On Advances In Engineering, Science And
Management (ICAESM -2012) March 2012.
[3] Alfieri, A., Bianco, A., Brandimarte, P., and Chiasserini, C.F.
"Maximizing system Lifetime in wireless sensor Networks",
European Journal of Operational Research, 2007.
[4] B. Paul, M. A. Martin, M.1. Showkat, and Z.
Rehman,"Optimal placement of base stations in a two tiered
wireless sensor network", WSEAS Transactions on
Communications, vol. 1, 2010.
[5] W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan,
An application-specific protocol architecture for wireless
micro sensor networks, IEEE Trans. Wireless Commun., vol.
1, no. 4, pp. 660660, Oct. 2002.
[6] O. Younis and S. Fahmy, Distributed clustering in ad-hoc
sensor networks: A hybrid, energy-efficient approach, in
IEEE Conf. Comput. Commun., pp. 366379, 2004.
[7] D. Gong, Y. Yang, and Z. Pan, Energy-efficient clustering in
lossy wireless sensor networks, J. Parallel Distrib. Comput.,
vol. 73, no. 9, pp. 13231336, Sep. 2013.

4
IJRITCC | December 2016, Available @ http://www.ijritcc.org
_______________________________________________________________________________________